for 474: forest inventory plot level metrics from lidar heights other plot measures sources of error...

FOR 474: Forest Inventory

Plot Level Metrics from Lidar

• Heights

• Other Plot Measures

• Sources of Error

Readings:

See Website

Plot Level Metrics: Getting at Canopy Heights

Heights are an Implicit Output of Lidar data

Calculating HeightsPlot Level Metrics: What is the Point Cloud Anyway?

Each distance from the plane to the surfaces is recorded:

Returns from surfaces further away from the sensor have a greater distance but a lower relative elevation than those “closer” returns

D1 D6

Calculating Heights

The point cloud shows the distances from the sensor to the surfaces. To be useful, we need to flip this up-side down world.

Plot Level Metrics: What is the Point Cloud Anyway?

Calculating Heights

We flip the data by subtracting the distance from the plane to the closer surfaces (Dc) from the distance from the plane to the furthest away surfaces (Df).

Plot Level Metrics: Getting at Canopy Heights

Dc Df Heights = Df- Dc

To get heights, subtract the “elevations” of the closer lidar points from the filtered “ground surface elevations” obtained from the Lidar DEM

Plot Level Metrics: Getting at Canopy Heights

This produces a point cloud where the DEM has a height of zero and the returns closer to the sensor have increasingly higher “heights”

Plot Level Metrics: Getting at Canopy Heights

z

0

Plot Level Metrics: Getting at Canopy Heights

When in forestry continuous surfaces are fit to the non-ground heights this is often called a “canopy height model”

Image source: H-E Anderson

Plot Level Metrics: Canopy Height Models

Image source: MJ Falkowski

Plot Level Metrics: Canopy Height Models

Interpolation Error:

The ground surface may be derived incorrectly due to insufficient ground returns at specific trees. Can occur in patches of high canopy cover or when sub-canopy features are present (seedlings, fuel buildup, etc)

• Trees too tall when ground surface is defined too low• Trees too short when ground surface is defined too high

Plot Level Metrics: Sources of Height Error

Plot Level Metrics: Sources of Height Error

Scale Error:

The ground surface may be derived incorrectly BUT have a consistent bias (up or down) due to insufficient ground returns across a series of trees.

This can also happen when the method to obtain the ground has been over-smoothed: i.e. too many returns deleted

Plot Level Metrics: Sources of Height Error

Tree Measurement Errors:

If too few returns are obtained per tree the maximum height may not be close to the actual tree height

In general Lidar will miss the tree top and will underestimate the true maximum tree height

Ideal Top Missed Tree Missed

Plot Level Metrics: Missing the Tree Top

The underestimation seen in Lidar is also a challenge in field tree measurements!!! When using clinometers, rangefinders, etc you may not be able to “see” the tree top!

Interaction between laser pulse (distance) and slope

This can be further influenced by scan angle

Plot Level Metrics: Sources of Height Error

Plot Level Metrics: Getting Maximum Tree Height

Tree Height

Popescu, S.C., Wynne, R.H. and Nelson, R.F. (2003.

Assume each local maximum in the canopy surface is a tree-top

Plot Level Metrics: Getting Maximum Tree Height

Valley Following

1. Assume each local maximum in the canopy surface is a tree-top

2. Apply contours to the canopy surface map

4. Find the local minimums surrounding each local maximum

5. Calculate Average N-S and E-W Diameter

Plot Level Metrics: Tree Crown Widths and Locations

Plot Level Metrics: Tree Crown Widths and Locations

10 m

5 m

Using a GIS:

Manually measure the width of each tree and delineate them into polygons

Using Allometric Equations:

1. Assume each local maximum in the canopy surface is a tree-top

2. Derive crown diameter from height relations:

cd = 2.56 * 0.14h

From:Falkowski, M.J., Smith, A.M.S., et al., (2006). Automated estimation of individual conifer tree height and crown diameter via Two-dimensional spatial wavelet analysis of lidar data, Canadian Journal of Remote Sensing, Vol. 32, No. 2, 153-161.

http://www.treesearch.fs.fed.us/pubs/24611

Plot Level Metrics: Tree Crown Widths and Locations

Using Automatic Methods

1. Convert each lidar canopy height model into a raster grid (via a GIS)

2. Use automated methods to ‘detect’ the location and crown width of each lidar tree

For more information see:Falkowski, M.J., Smith, A.M.S., et al., (2006). Automated estimation of individual conifer tree height and crown diameter via Two-dimensional spatial wavelet analysis of lidar data, Canadian Journal of Remote Sensing, Vol. 32, No. 2, 153-161.

http://www.treesearch.fs.fed.us/pubs/24611

Lidar Height Data

Plot Level Metrics: Tree Crown Widths and Locations

Crown Diameter

Using Automatic Methods

1. Convert each lidar canopy height model into a raster grid (via a GIS)

2. Use automated methods to ‘detect’ the location and crown width of each lidar tree

For more information see:Falkowski, M.J., Smith, A.M.S., et al., (2006). Automated estimation of individual conifer tree height and crown diameter via Two-dimensional spatial wavelet analysis of lidar data, Canadian Journal of Remote Sensing, Vol. 32, No. 2, 153-161.

http://www.treesearch.fs.fed.us/pubs/24611

Plot Level Metrics: Tree Crown Widths and Locations

Crown Diameter

Crown Base Height:

1. Convert each lidar canopy height model into a raster grid (via a GIS)

2. Use automated methods to ‘detect’ the location and crown width of each lidar tree

3. Within the crown diameter find the lowest height > than a set value (e.g. assume heights < 1m from trees: shrubs, rocks, etc)

Plot Level Metrics: Crown Base Height

Crown Diameter

Crown Bulk Density

1. Convert each lidar canopy height model into a raster grid (via a GIS)

2. Use automated methods to ‘detect’ the location and crown width of each lidar tree

3. Assume trees have a specific shape – cone, cylinder Volume

4. Use allometric equations via field measures to get foliar biomass

CBD = Foliar Biomass / Volume

Plot Level Metrics: Crown Bulk Density

Analysis of the Lidar data will be able to highlight trees above the canopy and importantly how tall the neighboring trees are.

Plot Level Metrics: Crown Class

What do you think the main limitation is?

Lidar can’t yet measure DBH directly:

Must model DBH from tree heights and crown widths OR use other allometric methods to directly get Biomass. See Week 6 readings.

Plot Level Metrics: Diameter at Breast Height

This creates a challenge as most Growth & Yield and Productivity models rely on a measure of DBH.

Therefore we need to develop “Lidar aware” allometric relationships!

FOR 474: Forest Inventory

Next Time …

Stand Level Metrics from Lidar